platform/upstream/llvm.git
5 years agogn build: Merge r370512
Nico Weber [Fri, 30 Aug 2019 20:06:44 +0000 (20:06 +0000)]
gn build: Merge r370512

llvm-svn: 370516

5 years ago[X86] Fix mul test cases in avx512-broadcast-unfold.ll to not get canonicalized to...
Craig Topper [Fri, 30 Aug 2019 20:04:23 +0000 (20:04 +0000)]
[X86] Fix mul test cases in avx512-broadcast-unfold.ll to not get canonicalized to fadd. Remove the fsub test cases which were also testing fadd.

Not sure how to prevent an fsub by constant getting turned into an fadd by negative constant.

llvm-svn: 370515

5 years ago[clang-tidy] Add llvm-prefer-register-over-unsigned to clang-tidy
Daniel Sanders [Fri, 30 Aug 2019 20:01:59 +0000 (20:01 +0000)]
[clang-tidy] Add llvm-prefer-register-over-unsigned to clang-tidy

Summary:
This clang-tidy check is looking for unsigned integer variables whose initializer
starts with an implicit cast from llvm::Register and changes the type of the
variable to llvm::Register (dropping the llvm:: where possible).

Reviewers: arsenm, bogner

Subscribers: jholewinski, MatzeB, qcolombet, dschuff, jyknight, dylanmckay, sdardis, nemanjai, jvesely, wdng, nhaehnle, mgorny, sbc100, jgravelle-google, kristof.beyls, hiraditya, aheejin, kbarton, fedor.sergeev, javed.absar, asb, rbar, johnrusso, simoncook, apazos, sabuasal, niosHD, jrtc27, MaskRay, zzheng, edward-jones, atanasyan, rogfer01, MartinMosbeck, brucehoult, the_o, tpr, PkmX, jocewei, jsji, Petar.Avramovic, asbirlea, Jim, s.egerton, cfe-commits, llvm-commits

Tags: #clang, #llvm

Differential Revision: https://reviews.llvm.org/D65919

llvm-svn: 370512

5 years ago[IFS][NFC] llvm-ifs: Fixing build errors for bots using GCC.
Puyan Lotfi [Fri, 30 Aug 2019 19:54:46 +0000 (19:54 +0000)]
[IFS][NFC] llvm-ifs: Fixing build errors for bots using GCC.

gcc produces the error:

error: specialization of
‘template<class T, class Enable> struct llvm::yaml::ScalarTraits’ in
different namespace

For all specializations outside of llvm::yaml. So I added llvm::yaml to these
specializations to fix the errors on the bots building with gcc (/usr/bin/c++).

llvm-svn: 370510

5 years ago[lld][WebAssembly] Fix spurious signature mismatch warnings
Sam Clegg [Fri, 30 Aug 2019 19:50:59 +0000 (19:50 +0000)]
[lld][WebAssembly] Fix spurious signature mismatch warnings

Summary:
This a follow up on: https://reviews.llvm.org/D62153

Handle the case where there are multiple object files that contain
undefined references to the same function.  We only generate a function
variant if the existing symbol is directly called.

See: https://github.com/emscripten-core/emscripten/issues/8995

Subscribers: dschuff, jgravelle-google, aheejin, sunfish, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D67015

llvm-svn: 370509

5 years ago[DFAPacketizer] Allow namespacing of automata per-itinerary
James Molloy [Fri, 30 Aug 2019 19:50:49 +0000 (19:50 +0000)]
[DFAPacketizer] Allow namespacing of automata per-itinerary

The Hexagon itineraries are cunningly crafted such that functional units between
itineraries do not clash. Because all itineraries are bundled into the same DFA,
a functional unit index clash would cause an incorrect DFA to be generated.

A workaround for this is to ensure all itineraries declare the universe of all
possible functional units, but this isn't ideal for three reasons:
  1) We only have a limited number of FUs we can encode in the packetizer, and
     using the universe causes us to hit the limit without care.
  2) Silent codegen faults are bad, and careful triage of the FU list shouldn't
     be required.
  3) Smooshing all itineraries into the same automaton allows combinations of
     instruction classes that cannot exist, which bloats the table.

A simple solution is to allow "namespacing" packetizers.

Differential Revision: https://reviews.llvm.org/D66940

llvm-svn: 370508

5 years ago[X86] Regenerate the test cases added in r370506.
Craig Topper [Fri, 30 Aug 2019 19:42:48 +0000 (19:42 +0000)]
[X86] Regenerate the test cases added in r370506.

Something weird happened with the v2i64/v2f64 test cases which
don't use broadcast. So they should already be hoisted, but
weren't in the version I submitted in r370506. This fixes that.
Not sure if something changed or I screwed up.

llvm-svn: 370507

5 years ago[X86] Add test caes for opportunities for machine LICM to unfold broadcasted constant...
Craig Topper [Fri, 30 Aug 2019 19:26:06 +0000 (19:26 +0000)]
[X86] Add test caes for opportunities for machine LICM to unfold broadcasted constant pool loads.

MachineLICM is able to unfold loads to move an invariant load out
a loop, but X86 infrastructure currently lacks the ability to do
this when avx512 embedded broadcasting is used.

This test adds examples for the basic float point operations,
add, mul, and, or, and xor.

llvm-svn: 370506

5 years ago[PowerPC][NFC] Avoid checking non-relevant .cfi instructions
Jinsong Ji [Fri, 30 Aug 2019 19:24:25 +0000 (19:24 +0000)]
[PowerPC][NFC] Avoid checking non-relevant .cfi instructions

Summary:
This is brought up in
https://reviews.llvm.org/D64662?id=209923#inline-599490

CFI information are non-relevant to quite some testcases,
we should get rid of checking them when its unecessary.

This patch avoid generating cfi info in testcases that are not
testing prolog/epilog or exception handling.

Reviewers: kbarton, hfinkel, nemanjai, #powerpc

Reviewed By: hfinkel

Subscribers: MaskRay, shchenz, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D67016

llvm-svn: 370505

5 years agoFix compilation warnings. NFC.
Michael Liao [Fri, 30 Aug 2019 19:23:28 +0000 (19:23 +0000)]
Fix compilation warnings. NFC.

llvm-svn: 370504

5 years ago[lldb][NFC] More tests for invalid register command invocations
Raphael Isemann [Fri, 30 Aug 2019 19:19:25 +0000 (19:19 +0000)]
[lldb][NFC] More tests for invalid register command invocations

llvm-svn: 370503

5 years agoMake `vector` unconditionally move elements when exceptions are disabled.
Eric Fiselier [Fri, 30 Aug 2019 19:01:03 +0000 (19:01 +0000)]
Make `vector` unconditionally move elements when exceptions are disabled.

Summary:
`std::vector<T>` is free choose between using copy or move operations when it needs to resize. The standard only candidates that the correct exception safety guarantees are provided. When exceptions are disabled these guarantees are trivially satisfied. Meaning vector is free to optimize it's implementation by moving instead of copying.

This patch makes `std::vector` unconditionally move elements when exceptions are disabled.

This optimization is conforming according to the current standard wording.

There are concerns that moving in `-fno-noexceptions`mode will be a surprise to users. For example, a user may be surprised to find their code is slower with exceptions enabled than it is disabled. I'm sympathetic to this surprised, but I don't think it should block this optimization.

Reviewers: mclow.lists, ldionne, rsmith

Reviewed By: ldionne

Subscribers: zoecarver, christof, dexonsmith, libcxx-commits

Tags: #libc

Differential Revision: https://reviews.llvm.org/D62228

llvm-svn: 370502

5 years agogn build: Merge r370500
Nico Weber [Fri, 30 Aug 2019 18:55:11 +0000 (18:55 +0000)]
gn build: Merge r370500

llvm-svn: 370501

5 years ago[MachinePipeliner] Separate schedule emission, NFC
James Molloy [Fri, 30 Aug 2019 18:49:50 +0000 (18:49 +0000)]
[MachinePipeliner] Separate schedule emission, NFC

This is the first stage in refactoring the pipeliner and making it more
accessible for backends to override and control. This separates the logic and
state required to *emit* a scheudule from the logic that *computes* and
validates a schedule.

This will enable (a) new schedule emitters and (b) new modulo scheduling
implementations to coexist.

NFC.

Differential Revision: https://reviews.llvm.org/D67006

llvm-svn: 370500

5 years ago[llvm-ifs][IFS] llvm Interface Stubs merging + object file generation tool.
Puyan Lotfi [Fri, 30 Aug 2019 18:26:05 +0000 (18:26 +0000)]
[llvm-ifs][IFS] llvm Interface Stubs merging + object file generation tool.

This tool merges interface stub files to produce a merged interface stub file
or a stub library. Currently it for stub library generation it can produce an
ELF .so stub file, or a TBD file (experimental). It will be used by the clang
-emit-interface-stubs compilation pipeline to merge and assemble the per-CU
stub files into a stub library.

The new IFS format is as follows:

--- !experimental-ifs-v1
IfsVersion:      1.0
Triple:          <llvm triple>
ObjectFileFormat: <ELF | TBD>
Symbols:
  _ZSymbolName: { Type: <type>, etc... }
...

Differential Revision: https://reviews.llvm.org/D66405

llvm-svn: 370499

5 years ago[DAGCombine] ReduceLoadWidth - remove duplicate SDLoc. NFCI.
Simon Pilgrim [Fri, 30 Aug 2019 18:19:02 +0000 (18:19 +0000)]
[DAGCombine] ReduceLoadWidth - remove duplicate SDLoc. NFCI.

SDLoc(N0) and SDLoc(cast<LoadSDNode>(N0)) should be equivalent.

llvm-svn: 370498

5 years ago[TargetLowering] SimplifyDemandedBits ADD/SUB/MUL - correctly inherit SDNodeFlags...
Simon Pilgrim [Fri, 30 Aug 2019 17:58:55 +0000 (17:58 +0000)]
[TargetLowering] SimplifyDemandedBits ADD/SUB/MUL - correctly inherit SDNodeFlags from the original node.

Just disable NSW/NUW flags. This matches what we're already doing for the other situations for these nodes, it was just missed for the demanded constant case.

Noticed by inspection - confirmed in offline discussion with @spatel. I've checked we have test coverage in the x86 extract-bits.ll and extract-lowbits.ll tests

llvm-svn: 370497

5 years agoGlobalISel: Fix missing pass dependency
Matt Arsenault [Fri, 30 Aug 2019 17:41:58 +0000 (17:41 +0000)]
GlobalISel: Fix missing pass dependency

llvm-svn: 370496

5 years ago[X86] Pass v32i16/v64i8 in zmm registers on KNL target.
Craig Topper [Fri, 30 Aug 2019 17:35:08 +0000 (17:35 +0000)]
[X86] Pass v32i16/v64i8 in zmm registers on KNL target.

gcc and icc pass these types in zmm registers in zmm registers.

This patch implements a quick hack to override the register
type before calling convention handling to one that is legal.
Longer term we might want to do something similar to 256-bit
integer registers on AVX1 where we just split all the operations.

Fixes PR42957

Differential Revision: https://reviews.llvm.org/D66708

llvm-svn: 370495

5 years ago[ValueTypes] Add v16f16 and v32f16 to EVT::getEVTString and Tablegen's getEnumName
Craig Topper [Fri, 30 Aug 2019 17:34:29 +0000 (17:34 +0000)]
[ValueTypes] Add v16f16 and v32f16 to EVT::getEVTString and Tablegen's getEnumName

Missed these when I hadded the enum entries

llvm-svn: 370494

5 years ago[clang-scan-deps] NFC, remove outdated implementation comment
Alex Lorenz [Fri, 30 Aug 2019 17:34:22 +0000 (17:34 +0000)]
[clang-scan-deps] NFC, remove outdated implementation comment

There's no need to purge symlinked entries in the FileManager,
as the new FileEntryRef API allows us to compute dependencies more
accurately when the FileManager is reused.

llvm-svn: 370493

5 years agogn build: Merge r370490
Nico Weber [Fri, 30 Aug 2019 17:30:08 +0000 (17:30 +0000)]
gn build: Merge r370490

llvm-svn: 370492

5 years ago[LLD] [COFF] Add a missing REQUIRES line to a recently added test. NFC.
Martin Storsjo [Fri, 30 Aug 2019 17:26:54 +0000 (17:26 +0000)]
[LLD] [COFF] Add a missing REQUIRES line to a recently added test. NFC.

This should fix failing buildbots like
http://lab.llvm.org:8011/builders/clang-cmake-aarch64-lld/builds/7180.

llvm-svn: 370491

5 years agoMemTag: unchecked load/store optimization.
Evgeniy Stepanov [Fri, 30 Aug 2019 17:23:02 +0000 (17:23 +0000)]
MemTag: unchecked load/store optimization.

Summary:
MTE allows memory access to bypass tag check iff the address argument
is [SP, #imm]. This change takes advantage of this to demote uses of
tagged addresses to regular FrameIndex operands, reducing register
pressure in large functions.

MO_TAGGED target flag is used to signal that the FrameIndex operand
refers to memory that might be tagged, and needs to be handled with
care. Such operand must be lowered to [SP, #imm] directly, without a
scratch register.

The transformation pass attempts to predict when the offset will be
out of range and disable the optimization.
AArch64RegisterInfo::eliminateFrameIndex has an escape hatch in case
this prediction has been wrong, but it is quite inefficient and should
be avoided.

Reviewers: pcc, vitalybuka, ostannard

Subscribers: mgorny, javed.absar, kristof.beyls, hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D66457

llvm-svn: 370490

5 years ago[DAGCombine] visitVSELECT - remove equivalent getValueType() call. NFCI.
Simon Pilgrim [Fri, 30 Aug 2019 17:21:20 +0000 (17:21 +0000)]
[DAGCombine] visitVSELECT - remove equivalent getValueType() call. NFCI.

llvm-svn: 370489

5 years agoFileManager: Remove ShouldCloseOpenFile argument from getBufferForFile, NFC
Duncan P. N. Exon Smith [Fri, 30 Aug 2019 16:56:26 +0000 (16:56 +0000)]
FileManager: Remove ShouldCloseOpenFile argument from getBufferForFile, NFC

Remove this dead code.  We always close it.

llvm-svn: 370488

5 years ago[lld-link] implement -start-lib and -end-lib
Bob Haarman [Fri, 30 Aug 2019 16:50:10 +0000 (16:50 +0000)]
[lld-link] implement -start-lib and -end-lib

Summary:
This implements -start-lib and -end-lib flags for lld-link, analogous
to the similarly named options in ld.lld. Object files after
-start-lib are included in the link only when needed to resolve
undefined symbols. The -end-lib flag goes back to the normal behavior
of always including object files in the link. This mimics the
semantics of static libraries, but without needing to actually create
the archive file.

Reviewers: ruiu, smeenai, MaskRay

Reviewed By: ruiu, MaskRay

Subscribers: akhuang, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D66848

llvm-svn: 370487

5 years ago[INSTRUCTIONS] Add support of const for getLoadStorePointerOperand() and
Whitney Tsang [Fri, 30 Aug 2019 16:41:35 +0000 (16:41 +0000)]
[INSTRUCTIONS] Add support of const for getLoadStorePointerOperand() and
getLoadStorePointerOperand().
Reviewer: hsaito, sebpop, reames, hfinkel, mkuper, bogner, haicheng,
arsenm, lattner, chandlerc, grosser, rengolin
Reviewed By: reames
Subscribers: wdng, llvm-commits, bmahjour
Tag: LLVM
Differential Revision: https://reviews.llvm.org/D66595

llvm-svn: 370486

5 years ago[Attributor] Fix: do not pretend to preserve the CFG
Johannes Doerfert [Fri, 30 Aug 2019 16:35:10 +0000 (16:35 +0000)]
[Attributor] Fix: do not pretend to preserve the CFG

llvm-svn: 370485

5 years ago[X86] Merge X86InstrInfo::loadRegFromAddr/storeRegToAddr into their only call site.
Craig Topper [Fri, 30 Aug 2019 16:05:57 +0000 (16:05 +0000)]
[X86] Merge X86InstrInfo::loadRegFromAddr/storeRegToAddr into their only call site.

I'm looking at unfolding broadcast loads on AVX512 which will
require refactoring this code to select broadcast opcodes instead
of regular load/stores in some cases. Merging them to avoid
further complicating their interfaces.

llvm-svn: 370484

5 years ago[lit] Fix my earlier bogus fix to not set DYLD_LIBRARY_PATH with Asan.
Jonas Devlieghere [Fri, 30 Aug 2019 15:56:14 +0000 (15:56 +0000)]
[lit] Fix my earlier bogus fix to not set DYLD_LIBRARY_PATH with Asan.

My follow-up commit to mess with DYLD_LIBRARY_PATH was bogus for two
reasons:

 - The condition was inverted.
 - We were checking the OS's environment, instead of the config's.

Two wrongs don't make a right, but the second mistake meant that the
sanitizer bot passed.

llvm-svn: 370483

5 years ago[clangd] Add highlighting for macro expansions.
Johan Vikstrom [Fri, 30 Aug 2019 15:47:27 +0000 (15:47 +0000)]
[clangd] Add highlighting for macro expansions.

Summary: https://github.com/clangd/clangd/issues/134

Reviewers: hokein, ilya-biryukov

Subscribers: MaskRay, jkorous, arphaman, kadircet, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D66995

llvm-svn: 370482

5 years agoRevert [Clang Interpreter] Initial patch for the constexpr interpreter
Nandor Licker [Fri, 30 Aug 2019 15:41:45 +0000 (15:41 +0000)]
Revert [Clang Interpreter] Initial patch for the constexpr interpreter

This reverts r370476 (git commit a5590950549719d0d9ea69ed164b0c8c0f4e02e6)

llvm-svn: 370481

5 years ago[Attributor] Use existing function information for the call site
Johannes Doerfert [Fri, 30 Aug 2019 15:24:52 +0000 (15:24 +0000)]
[Attributor] Use existing function information for the call site

Summary:
Instead of recomputing information for call sites we now use the
function information directly. This is always valid and once we have
call site specific information we can improve here.

This patch also bootstraps attributes that are created on-demand through
an initial update call. Information that is known will then directly be
available in the new attribute without causing an iteration delay.

The tests show how this improves the iteration count.

Reviewers: sstefan1, uenoku

Subscribers: hiraditya, bollu, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D66781

llvm-svn: 370480

5 years ago[Attributor] Manifest load/store alignment generally
Johannes Doerfert [Fri, 30 Aug 2019 15:22:28 +0000 (15:22 +0000)]
[Attributor] Manifest load/store alignment generally

Summary:
Any pointer could have load/store users not only floating ones so we
move the manifest logic for alignment into the AAAlignImpl class.

Reviewers: uenoku, sstefan1

Subscribers: hiraditya, bollu, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D66922

llvm-svn: 370479

5 years ago[DAGCombine] visitVSELECT - remove duplicate getOperand calls. NFCI.
Simon Pilgrim [Fri, 30 Aug 2019 15:17:37 +0000 (15:17 +0000)]
[DAGCombine] visitVSELECT - remove duplicate getOperand calls. NFCI.

llvm-svn: 370478

5 years ago[Clang Interpreter] Initial patch for the constexpr interpreter
Nandor Licker [Fri, 30 Aug 2019 15:02:09 +0000 (15:02 +0000)]
[Clang Interpreter] Initial patch for the constexpr interpreter

Summary:
This patch introduces the skeleton of the constexpr interpreter,
capable of evaluating a simple constexpr functions consisting of
if statements. The interpreter is described in more detail in the
RFC. Further patches will add more features.

Reviewers: Bigcheese, jfb, rsmith

Subscribers: bruno, uenoku, ldionne, Tyker, thegameg, tschuett, dexonsmith, mgorny, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D64146

llvm-svn: 370476

5 years ago[InstCombine][AMDGPU] Simplify tbuffer loads
Piotr Sobczak [Fri, 30 Aug 2019 14:20:04 +0000 (14:20 +0000)]
[InstCombine][AMDGPU] Simplify tbuffer loads

Summary: Add missing tbuffer loads intrinsics in SimplifyDemandedVectorElts.

Reviewers: arsenm, nhaehnle

Reviewed By: arsenm

Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D66926

llvm-svn: 370475

5 years ago[llvm-nm] Small fix to Exected<StringRef>
Sid Manning [Fri, 30 Aug 2019 14:12:04 +0000 (14:12 +0000)]
[llvm-nm] Small fix to Exected<StringRef>

Differential Revision: https://reviews.llvm.org/D66976

llvm-svn: 370474

5 years ago[clangd] Added highlighting for structured bindings.
Johan Vikstrom [Fri, 30 Aug 2019 14:07:05 +0000 (14:07 +0000)]
[clangd] Added highlighting for structured bindings.

Summary: Structured bindings are in a BindingDecl. The decl the declRefExpr points to are the BindingDecls. So this adds an additional if statement in the addToken function to highlight them.

Reviewers: hokein, ilya-biryukov

Subscribers: MaskRay, jkorous, arphaman, kadircet, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D66738

llvm-svn: 370473

5 years ago[yaml2obj][obj2yaml] - Use a single "Other" field instead of "Other", "Visibility...
George Rimar [Fri, 30 Aug 2019 13:39:22 +0000 (13:39 +0000)]
[yaml2obj][obj2yaml] - Use a single "Other" field instead of "Other", "Visibility" and "StOther".

Currenly we can encode the 'st_other' field of symbol using 3 fields.
'Visibility' is used to encode STV_* values.
'Other' is used to encode everything except the visibility, but it can't handle arbitrary values.
'StOther' is used to encode arbitrary values when 'Visibility'/'Other' are not helpfull enough.

'st_other' field is used to encode symbol visibility and platform-dependent
flags and values. Problem to encode it is that it consists of Visibility part (STV_* values)
which are enumeration values and the Other part, which is different and inconsistent.

For MIPS the Other part contains flags for all STO_MIPS_* values except STO_MIPS_MIPS16.
(Like comment in ELFDumper says: "Someones in their infinite wisdom decided to make
STO_MIPS_MIPS16 flag overlapped with other ST_MIPS_xxx flags."...)

And for PPC64 the Other part might actually encode any value.

This patch implements custom logic for handling the st_other and removes
'Visibility' and 'StOther' fields.

Here is an example of a new YAML style this patch allows:

- Name:  foo
  Other: [ 0x4 ]
- Name:  bar
  Other: [ STV_PROTECTED, 4 ]
- Name:  zed
  Other: [ STV_PROTECTED, STO_MIPS_OPTIONAL, 0xf8 ]

Differential revision: https://reviews.llvm.org/D66886

llvm-svn: 370472

5 years ago[DAGCombine] visitVSELECT - use getShiftAmountTy for shift amounts.
Simon Pilgrim [Fri, 30 Aug 2019 13:30:37 +0000 (13:30 +0000)]
[DAGCombine] visitVSELECT - use getShiftAmountTy for shift amounts.

llvm-svn: 370471

5 years ago[DAGCombine] visitMULHS - use getScalarValueSizeInBits() to make safe for vector...
Simon Pilgrim [Fri, 30 Aug 2019 12:22:06 +0000 (12:22 +0000)]
[DAGCombine] visitMULHS - use getScalarValueSizeInBits() to make safe for vector types.

This is hidden behind a (scalar-only) isOneConstant(N1) check at the moment, but once we get around to adding vector support we need to ensure we're dealing with the scalar bitwidth, not the total.

llvm-svn: 370468

5 years ago[mips] Merge common checkings under the same check prefix. NFC
Simon Atanasyan [Fri, 30 Aug 2019 12:15:12 +0000 (12:15 +0000)]
[mips] Merge common checkings under the same check prefix. NFC

llvm-svn: 370467

5 years ago[RISCV] Fix a couple of tests' CHECKs
Luis Marques [Fri, 30 Aug 2019 12:11:47 +0000 (12:11 +0000)]
[RISCV] Fix a couple of tests' CHECKs

llvm-svn: 370466

5 years agoRemove an extra ";", NFC.
Haojian Wu [Fri, 30 Aug 2019 12:09:31 +0000 (12:09 +0000)]
Remove an extra ";", NFC.

llvm-svn: 370465

5 years ago[X86] Add tests for rotate matching. NFC
Amaury Sechet [Fri, 30 Aug 2019 11:35:28 +0000 (11:35 +0000)]
[X86] Add tests for rotate matching. NFC

llvm-svn: 370464

5 years ago[CodeGen] Introduce MachineBasicBlock::replacePhiUsesWith helper and use it. NFC
Bjorn Pettersson [Fri, 30 Aug 2019 11:23:10 +0000 (11:23 +0000)]
[CodeGen] Introduce MachineBasicBlock::replacePhiUsesWith helper and use it. NFC

Summary:
Found a couple of places in the code where all the PHI nodes
of a MBB is updated, replacing references to one MBB by
reference to another MBB instead.

This patch simply refactors the code to use a common helper
(MachineBasicBlock::replacePhiUsesWith) for such PHI node
updates.

Reviewers: t.p.northover, arsenm, uabelho

Subscribers: wdng, hiraditya, jsji, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D66750

llvm-svn: 370463

5 years ago[dotest] Finish removing -q
Pavel Labath [Fri, 30 Aug 2019 11:02:58 +0000 (11:02 +0000)]
[dotest] Finish removing -q

One usage of this option remained, and caused dotest to error out if one
happened to pass the -v flag.

llvm-svn: 370462

5 years ago[ASTImporter] Do not look up lambda classes
Gabor Marton [Fri, 30 Aug 2019 10:55:41 +0000 (10:55 +0000)]
[ASTImporter] Do not look up lambda classes

Summary:
Consider this code:
```
      void f() {
        auto L0 = [](){};
        auto L1 = [](){};
      }

```
First we import `L0` then `L1`. Currently we end up having only one
CXXRecordDecl for the two different lambdas. And that is a problem if
the body of their op() is different. This happens because when we import
`L1` then lookup finds the existing `L0` and since they are structurally
equivalent we just map the imported L0 to be the counterpart of L1.

We have the same problem in this case:
```
      template <typename F0, typename F1>
      void f(F0 L0 = [](){}, F1 L1 = [](){}) {}

```

In StructuralEquivalenceContext we could distinquish lambdas only by
their source location in these cases. But we the lambdas are actually
structrually equivalent they differn only by the source location.

Thus, the  solution is to disable lookup completely if the decl in
the "from" context is a lambda.
However, that could have other problems: what if the lambda is defined
in a header file and included in several TUs? I think we'd have as many
duplicates as many includes we have. I think we could live with that,
because the lambda classes are TU local anyway, we cannot just access
them from another TU.

Reviewers: a_sidorin, a.sidorin, shafik

Subscribers: rnkovacs, dkrupp, Szelethus, gamesh411, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D66348

llvm-svn: 370461

5 years ago[DAGCombine] visitMULHS/visitMULHU - isBuildVectorAllZeros doesn't mean node is all...
Simon Pilgrim [Fri, 30 Aug 2019 10:42:14 +0000 (10:42 +0000)]
[DAGCombine] visitMULHS/visitMULHU - isBuildVectorAllZeros doesn't mean node is all zeros

Return a proper zero vector, just in case some elements are undef.

Noticed by inspection after dealing with a similar issue in PR43159.

llvm-svn: 370460

5 years agoFix Wdocumentation warning. NFCI.
Simon Pilgrim [Fri, 30 Aug 2019 10:25:52 +0000 (10:25 +0000)]
Fix Wdocumentation warning. NFCI.

llvm-svn: 370459

5 years ago[llvm-objcopy] Allow the visibility of symbols created by --binary and
Chris Jackson [Fri, 30 Aug 2019 10:17:16 +0000 (10:17 +0000)]
[llvm-objcopy] Allow the visibility of symbols created by --binary and
--add-symbol to be specified with --new-symbol-visibility

llvm-svn: 370458

5 years ago[ASTImporter] Propagate errors during import of overridden methods.
Balazs Keri [Fri, 30 Aug 2019 10:12:14 +0000 (10:12 +0000)]
[ASTImporter] Propagate errors during import of overridden methods.

Summary:
If importing overridden methods fails for a method it can be seen
incorrectly as non-virtual. To avoid this inconsistency the method
is marked with import error to avoid later use of it.

Reviewers: martong, a.sidorin, shafik, a_sidorin

Reviewed By: martong, shafik

Subscribers: rnkovacs, dkrupp, Szelethus, gamesh411, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D66933

llvm-svn: 370457

5 years ago[Attributor] Implement AANoAliasCallSiteArgument initialization
Hideto Ueno [Fri, 30 Aug 2019 10:00:32 +0000 (10:00 +0000)]
[Attributor] Implement AANoAliasCallSiteArgument initialization

Summary: This patch adds an appropriate `initialize` method for `AANoAliasCallSiteArgument`.

Reviewers: jdoerfert, sstefan1

Reviewed By: jdoerfert

Subscribers: hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D66927

llvm-svn: 370456

5 years ago[Clangd] ExtractFunction Added checks for broken control flow
Shaurya Gupta [Fri, 30 Aug 2019 09:57:56 +0000 (09:57 +0000)]
[Clangd] ExtractFunction Added checks for broken control flow

Summary:
- Added checks for broken control flow
- Added unittests

Reviewers: sammccall, kadircet

Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D66732

llvm-svn: 370455

5 years ago[LoopIdiomRecognize] BCmp loop idiom recognition
Roman Lebedev [Fri, 30 Aug 2019 09:51:23 +0000 (09:51 +0000)]
[LoopIdiomRecognize] BCmp loop idiom recognition

Summary:
@mclow.lists brought up this issue up in IRC.
It is a reasonably common problem to compare some two values for equality.
Those may be just some integers, strings or arrays of integers.

In C, there is `memcmp()`, `bcmp()` functions.
In C++, there exists `std::equal()` algorithm.
One can also write that function manually.

libstdc++'s `std::equal()` is specialized to directly call `memcmp()` for
various types, but not `std::byte` from C++2a. https://godbolt.org/z/mx2ejJ

libc++ does not do anything like that, it simply relies on simple C++'s
`operator==()`. https://godbolt.org/z/er0Zwf (GOOD!)

So likely, there exists a certain performance opportunities.
Let's compare performance of naive `std::equal()` (no `memcmp()`) with one that
is using `memcmp()` (in this case, compiled with modified compiler). {F8768213}

```
#include <algorithm>
#include <cmath>
#include <cstdint>
#include <iterator>
#include <limits>
#include <random>
#include <type_traits>
#include <utility>
#include <vector>

#include "benchmark/benchmark.h"

template <class T>
bool equal(T* a, T* a_end, T* b) noexcept {
  for (; a != a_end; ++a, ++b) {
    if (*a != *b) return false;
  }
  return true;
}

template <typename T>
std::vector<T> getVectorOfRandomNumbers(size_t count) {
  std::random_device rd;
  std::mt19937 gen(rd());
  std::uniform_int_distribution<T> dis(std::numeric_limits<T>::min(),
                                       std::numeric_limits<T>::max());
  std::vector<T> v;
  v.reserve(count);
  std::generate_n(std::back_inserter(v), count,
                  [&dis, &gen]() { return dis(gen); });
  assert(v.size() == count);
  return v;
}

struct Identical {
  template <typename T>
  static std::pair<std::vector<T>, std::vector<T>> Gen(size_t count) {
    auto Tmp = getVectorOfRandomNumbers<T>(count);
    return std::make_pair(Tmp, std::move(Tmp));
  }
};

struct InequalHalfway {
  template <typename T>
  static std::pair<std::vector<T>, std::vector<T>> Gen(size_t count) {
    auto V0 = getVectorOfRandomNumbers<T>(count);
    auto V1 = V0;
    V1[V1.size() / size_t(2)]++;  // just change the value.
    return std::make_pair(std::move(V0), std::move(V1));
  }
};

template <class T, class Gen>
void BM_bcmp(benchmark::State& state) {
  const size_t Length = state.range(0);

  const std::pair<std::vector<T>, std::vector<T>> Data =
      Gen::template Gen<T>(Length);
  const std::vector<T>& a = Data.first;
  const std::vector<T>& b = Data.second;
  assert(a.size() == Length && b.size() == a.size());

  benchmark::ClobberMemory();
  benchmark::DoNotOptimize(a);
  benchmark::DoNotOptimize(a.data());
  benchmark::DoNotOptimize(b);
  benchmark::DoNotOptimize(b.data());

  for (auto _ : state) {
    const bool is_equal = equal(a.data(), a.data() + a.size(), b.data());
    benchmark::DoNotOptimize(is_equal);
  }
  state.SetComplexityN(Length);
  state.counters["eltcnt"] =
      benchmark::Counter(Length, benchmark::Counter::kIsIterationInvariant);
  state.counters["eltcnt/sec"] =
      benchmark::Counter(Length, benchmark::Counter::kIsIterationInvariantRate);
  const size_t BytesRead = 2 * sizeof(T) * Length;
  state.counters["bytes_read/iteration"] =
      benchmark::Counter(BytesRead, benchmark::Counter::kDefaults,
                         benchmark::Counter::OneK::kIs1024);
  state.counters["bytes_read/sec"] = benchmark::Counter(
      BytesRead, benchmark::Counter::kIsIterationInvariantRate,
      benchmark::Counter::OneK::kIs1024);
}

template <typename T>
static void CustomArguments(benchmark::internal::Benchmark* b) {
  const size_t L2SizeBytes = []() {
    for (const benchmark::CPUInfo::CacheInfo& I :
         benchmark::CPUInfo::Get().caches) {
      if (I.level == 2) return I.size;
    }
    return 0;
  }();
  // What is the largest range we can check to always fit within given L2 cache?
  const size_t MaxLen = L2SizeBytes / /*total bufs*/ 2 /
                        /*maximal elt size*/ sizeof(T) / /*safety margin*/ 2;
  b->RangeMultiplier(2)->Range(1, MaxLen)->Complexity(benchmark::oN);
}

BENCHMARK_TEMPLATE(BM_bcmp, uint8_t, Identical)
    ->Apply(CustomArguments<uint8_t>);
BENCHMARK_TEMPLATE(BM_bcmp, uint16_t, Identical)
    ->Apply(CustomArguments<uint16_t>);
BENCHMARK_TEMPLATE(BM_bcmp, uint32_t, Identical)
    ->Apply(CustomArguments<uint32_t>);
BENCHMARK_TEMPLATE(BM_bcmp, uint64_t, Identical)
    ->Apply(CustomArguments<uint64_t>);

BENCHMARK_TEMPLATE(BM_bcmp, uint8_t, InequalHalfway)
    ->Apply(CustomArguments<uint8_t>);
BENCHMARK_TEMPLATE(BM_bcmp, uint16_t, InequalHalfway)
    ->Apply(CustomArguments<uint16_t>);
BENCHMARK_TEMPLATE(BM_bcmp, uint32_t, InequalHalfway)
    ->Apply(CustomArguments<uint32_t>);
BENCHMARK_TEMPLATE(BM_bcmp, uint64_t, InequalHalfway)
    ->Apply(CustomArguments<uint64_t>);
```
{F8768210}
```
$ ~/src/googlebenchmark/tools/compare.py --no-utest benchmarks build-{old,new}/test/llvm-bcmp-bench
RUNNING: build-old/test/llvm-bcmp-bench --benchmark_out=/tmp/tmpb6PEUx
2019-04-25 21:17:11
Running build-old/test/llvm-bcmp-bench
Run on (8 X 4000 MHz CPU s)
CPU Caches:
  L1 Data 16K (x8)
  L1 Instruction 64K (x4)
  L2 Unified 2048K (x4)
  L3 Unified 8192K (x1)
Load Average: 0.65, 3.90, 4.14
---------------------------------------------------------------------------------------------------
Benchmark                                         Time             CPU   Iterations UserCounters...
---------------------------------------------------------------------------------------------------
<...>
BM_bcmp<uint8_t, Identical>/512000           432131 ns       432101 ns         1613 bytes_read/iteration=1000k bytes_read/sec=2.20706G/s eltcnt=825.856M eltcnt/sec=1.18491G/s
BM_bcmp<uint8_t, Identical>_BigO               0.86 N          0.86 N
BM_bcmp<uint8_t, Identical>_RMS                   8 %             8 %
<...>
BM_bcmp<uint16_t, Identical>/256000          161408 ns       161409 ns         4027 bytes_read/iteration=1000k bytes_read/sec=5.90843G/s eltcnt=1030.91M eltcnt/sec=1.58603G/s
BM_bcmp<uint16_t, Identical>_BigO              0.67 N          0.67 N
BM_bcmp<uint16_t, Identical>_RMS                 25 %            25 %
<...>
BM_bcmp<uint32_t, Identical>/128000           81497 ns        81488 ns         8415 bytes_read/iteration=1000k bytes_read/sec=11.7032G/s eltcnt=1077.12M eltcnt/sec=1.57078G/s
BM_bcmp<uint32_t, Identical>_BigO              0.71 N          0.71 N
BM_bcmp<uint32_t, Identical>_RMS                 42 %            42 %
<...>
BM_bcmp<uint64_t, Identical>/64000            50138 ns        50138 ns        10909 bytes_read/iteration=1000k bytes_read/sec=19.0209G/s eltcnt=698.176M eltcnt/sec=1.27647G/s
BM_bcmp<uint64_t, Identical>_BigO              0.84 N          0.84 N
BM_bcmp<uint64_t, Identical>_RMS                 27 %            27 %
<...>
BM_bcmp<uint8_t, InequalHalfway>/512000      192405 ns       192392 ns         3638 bytes_read/iteration=1000k bytes_read/sec=4.95694G/s eltcnt=1.86266G eltcnt/sec=2.66124G/s
BM_bcmp<uint8_t, InequalHalfway>_BigO          0.38 N          0.38 N
BM_bcmp<uint8_t, InequalHalfway>_RMS              3 %             3 %
<...>
BM_bcmp<uint16_t, InequalHalfway>/256000     127858 ns       127860 ns         5477 bytes_read/iteration=1000k bytes_read/sec=7.45873G/s eltcnt=1.40211G eltcnt/sec=2.00219G/s
BM_bcmp<uint16_t, InequalHalfway>_BigO         0.50 N          0.50 N
BM_bcmp<uint16_t, InequalHalfway>_RMS             0 %             0 %
<...>
BM_bcmp<uint32_t, InequalHalfway>/128000      49140 ns        49140 ns        14281 bytes_read/iteration=1000k bytes_read/sec=19.4072G/s eltcnt=1.82797G eltcnt/sec=2.60478G/s
BM_bcmp<uint32_t, InequalHalfway>_BigO         0.40 N          0.40 N
BM_bcmp<uint32_t, InequalHalfway>_RMS            18 %            18 %
<...>
BM_bcmp<uint64_t, InequalHalfway>/64000       32101 ns        32099 ns        21786 bytes_read/iteration=1000k bytes_read/sec=29.7101G/s eltcnt=1.3943G eltcnt/sec=1.99381G/s
BM_bcmp<uint64_t, InequalHalfway>_BigO         0.50 N          0.50 N
BM_bcmp<uint64_t, InequalHalfway>_RMS             1 %             1 %
RUNNING: build-new/test/llvm-bcmp-bench --benchmark_out=/tmp/tmpQ46PP0
2019-04-25 21:19:29
Running build-new/test/llvm-bcmp-bench
Run on (8 X 4000 MHz CPU s)
CPU Caches:
  L1 Data 16K (x8)
  L1 Instruction 64K (x4)
  L2 Unified 2048K (x4)
  L3 Unified 8192K (x1)
Load Average: 1.01, 2.85, 3.71
---------------------------------------------------------------------------------------------------
Benchmark                                         Time             CPU   Iterations UserCounters...
---------------------------------------------------------------------------------------------------
<...>
BM_bcmp<uint8_t, Identical>/512000            18593 ns        18590 ns        37565 bytes_read/iteration=1000k bytes_read/sec=51.2991G/s eltcnt=19.2333G eltcnt/sec=27.541G/s
BM_bcmp<uint8_t, Identical>_BigO               0.04 N          0.04 N
BM_bcmp<uint8_t, Identical>_RMS                  37 %            37 %
<...>
BM_bcmp<uint16_t, Identical>/256000           18950 ns        18948 ns        37223 bytes_read/iteration=1000k bytes_read/sec=50.3324G/s eltcnt=9.52909G eltcnt/sec=13.511G/s
BM_bcmp<uint16_t, Identical>_BigO              0.08 N          0.08 N
BM_bcmp<uint16_t, Identical>_RMS                 34 %            34 %
<...>
BM_bcmp<uint32_t, Identical>/128000           18627 ns        18627 ns        37895 bytes_read/iteration=1000k bytes_read/sec=51.198G/s eltcnt=4.85056G eltcnt/sec=6.87168G/s
BM_bcmp<uint32_t, Identical>_BigO              0.16 N          0.16 N
BM_bcmp<uint32_t, Identical>_RMS                 35 %            35 %
<...>
BM_bcmp<uint64_t, Identical>/64000            18855 ns        18855 ns        37458 bytes_read/iteration=1000k bytes_read/sec=50.5791G/s eltcnt=2.39731G eltcnt/sec=3.3943G/s
BM_bcmp<uint64_t, Identical>_BigO              0.32 N          0.32 N
BM_bcmp<uint64_t, Identical>_RMS                 33 %            33 %
<...>
BM_bcmp<uint8_t, InequalHalfway>/512000        9570 ns         9569 ns        73500 bytes_read/iteration=1000k bytes_read/sec=99.6601G/s eltcnt=37.632G eltcnt/sec=53.5046G/s
BM_bcmp<uint8_t, InequalHalfway>_BigO          0.02 N          0.02 N
BM_bcmp<uint8_t, InequalHalfway>_RMS             29 %            29 %
<...>
BM_bcmp<uint16_t, InequalHalfway>/256000       9547 ns         9547 ns        74343 bytes_read/iteration=1000k bytes_read/sec=99.8971G/s eltcnt=19.0318G eltcnt/sec=26.8159G/s
BM_bcmp<uint16_t, InequalHalfway>_BigO         0.04 N          0.04 N
BM_bcmp<uint16_t, InequalHalfway>_RMS            29 %            29 %
<...>
BM_bcmp<uint32_t, InequalHalfway>/128000       9396 ns         9394 ns        73521 bytes_read/iteration=1000k bytes_read/sec=101.518G/s eltcnt=9.41069G eltcnt/sec=13.6255G/s
BM_bcmp<uint32_t, InequalHalfway>_BigO         0.08 N          0.08 N
BM_bcmp<uint32_t, InequalHalfway>_RMS            30 %            30 %
<...>
BM_bcmp<uint64_t, InequalHalfway>/64000        9499 ns         9498 ns        73802 bytes_read/iteration=1000k bytes_read/sec=100.405G/s eltcnt=4.72333G eltcnt/sec=6.73808G/s
BM_bcmp<uint64_t, InequalHalfway>_BigO         0.16 N          0.16 N
BM_bcmp<uint64_t, InequalHalfway>_RMS            28 %            28 %
Comparing build-old/test/llvm-bcmp-bench to build-new/test/llvm-bcmp-bench
Benchmark                                                  Time             CPU      Time Old      Time New       CPU Old       CPU New
---------------------------------------------------------------------------------------------------------------------------------------
<...>
BM_bcmp<uint8_t, Identical>/512000                      -0.9570         -0.9570        432131         18593        432101         18590
<...>
BM_bcmp<uint16_t, Identical>/256000                     -0.8826         -0.8826        161408         18950        161409         18948
<...>
BM_bcmp<uint32_t, Identical>/128000                     -0.7714         -0.7714         81497         18627         81488         18627
<...>
BM_bcmp<uint64_t, Identical>/64000                      -0.6239         -0.6239         50138         18855         50138         18855
<...>
BM_bcmp<uint8_t, InequalHalfway>/512000                 -0.9503         -0.9503        192405          9570        192392          9569
<...>
BM_bcmp<uint16_t, InequalHalfway>/256000                -0.9253         -0.9253        127858          9547        127860          9547
<...>
BM_bcmp<uint32_t, InequalHalfway>/128000                -0.8088         -0.8088         49140          9396         49140          9394
<...>
BM_bcmp<uint64_t, InequalHalfway>/64000                 -0.7041         -0.7041         32101          9499         32099          9498
```

What can we tell from the benchmark?
* Performance of naive equality check somewhat improves with element size,
  maxing out at eltcnt/sec=1.58603G/s for uint16_t, or bytes_read/sec=19.0209G/s
  for uint64_t. I think, that instability implies performance problems.
* Performance of `memcmp()`-aware benchmark always maxes out at around
  bytes_read/sec=51.2991G/s for every type. That is 2.6x the throughput of the
  naive variant!
* eltcnt/sec metric for the `memcmp()`-aware benchmark maxes out at
  eltcnt/sec=27.541G/s for uint8_t (was: eltcnt/sec=1.18491G/s, so 24x) and
  linearly decreases with element size.
  For uint64_t, it's ~4x+ the elements/second.
* The call obvious is more pricey than the loop, with small element count.
  As it can be seen from the full output {F8768210}, the `memcmp()` is almost
  universally worse, independent of the element size (and thus buffer size) when
  element count is less than 8.

So all in all, bcmp idiom does indeed pose untapped performance headroom.
This diff does implement said idiom recognition. I think a reasonable test
coverage is present, but do tell if there is anything obvious missing.

Now, quality. This does succeed to build and pass the test-suite, at least
without any non-bundled elements. {F8768216} {F8768217}
This transform fires 91 times:
```
$ /build/test-suite/utils/compare.py -m loop-idiom.NumBCmp result-new.json
Tests: 1149
Metric: loop-idiom.NumBCmp

Program                                         result-new

MultiSourc...Benchmarks/7zip/7zip-benchmark    79.00
MultiSource/Applications/d/make_dparser         3.00
SingleSource/UnitTests/vla                      2.00
MultiSource/Applications/Burg/burg              1.00
MultiSourc.../Applications/JM/lencod/lencod     1.00
MultiSource/Applications/lemon/lemon            1.00
MultiSource/Benchmarks/Bullet/bullet            1.00
MultiSourc...e/Benchmarks/MallocBench/gs/gs     1.00
MultiSourc...gs-C/TimberWolfMC/timberwolfmc     1.00
MultiSourc...Prolangs-C/simulator/simulator     1.00
```
The size changes are:
I'm not sure what's going on with SingleSource/UnitTests/vla.test yet, did not look.
```
$ /build/test-suite/utils/compare.py -m size..text result-{old,new}.json --filter-hash
Tests: 1149
Same hash: 907 (filtered out)
Remaining: 242
Metric: size..text

Program                                        result-old result-new diff
test-suite...ingleSource/UnitTests/vla.test   753.00     833.00     10.6%
test-suite...marks/7zip/7zip-benchmark.test   1001697.00 966657.00  -3.5%
test-suite...ngs-C/simulator/simulator.test   32369.00   32321.00   -0.1%
test-suite...plications/d/make_dparser.test   89585.00   89505.00   -0.1%
test-suite...ce/Applications/Burg/burg.test   40817.00   40785.00   -0.1%
test-suite.../Applications/lemon/lemon.test   47281.00   47249.00   -0.1%
test-suite...TimberWolfMC/timberwolfmc.test   250065.00  250113.00   0.0%
test-suite...chmarks/MallocBench/gs/gs.test   149889.00  149873.00  -0.0%
test-suite...ications/JM/lencod/lencod.test   769585.00  769569.00  -0.0%
test-suite.../Benchmarks/Bullet/bullet.test   770049.00  770049.00   0.0%
test-suite...HMARK_ANISTROPIC_DIFFUSION/128    NaN        NaN        nan%
test-suite...HMARK_ANISTROPIC_DIFFUSION/256    NaN        NaN        nan%
test-suite...CHMARK_ANISTROPIC_DIFFUSION/64    NaN        NaN        nan%
test-suite...CHMARK_ANISTROPIC_DIFFUSION/32    NaN        NaN        nan%
test-suite...ENCHMARK_BILATERAL_FILTER/64/4    NaN        NaN        nan%
Geomean difference                                                   nan%
         result-old    result-new       diff
count  1.000000e+01  10.00000      10.000000
mean   3.152090e+05  311695.40000  0.006749
std    3.790398e+05  372091.42232  0.036605
min    7.530000e+02  833.00000    -0.034981
25%    4.243300e+04  42401.00000  -0.000866
50%    1.197370e+05  119689.00000 -0.000392
75%    6.397050e+05  639705.00000 -0.000005
max    1.001697e+06  966657.00000  0.106242
```

I don't have timings though.

And now to the code. The basic idea is to completely replace the whole loop.
If we can't fully kill it, don't transform.
I have left one or two comments in the code, so hopefully it can be understood.

Also, there is a few TODO's that i have left for follow-ups:
* widening of `memcmp()`/`bcmp()`
* step smaller than the comparison size
* Metadata propagation
* more than two blocks as long as there is still a single backedge?
* ???

Reviewers: reames, fhahn, mkazantsev, chandlerc, craig.topper, courbet

Reviewed By: courbet

Subscribers: hiraditya, xbolva00, nikic, jfb, gchatelet, courbet, llvm-commits, mclow.lists

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D61144

llvm-svn: 370454

5 years ago[NFC] SCEVExpander: add SetCurrentDebugLocation() / getCurrentDebugLocation() wrappers
Roman Lebedev [Fri, 30 Aug 2019 09:51:02 +0000 (09:51 +0000)]
[NFC] SCEVExpander: add SetCurrentDebugLocation() / getCurrentDebugLocation() wrappers

Summary:
The internal `Builder` is private, which means there is
currently no way to set the debuginfo locations for `SCEVExpander`.
This only adds the wrappers, but does not use them anywhere.

Reviewers: mkazantsev, sanjoy, gberry, jyknight, dneilson

Reviewed By: sanjoy

Subscribers: javed.absar, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D61007

llvm-svn: 370453

5 years ago[clangd] Collecting main file macro expansion locations in ParsedAST.
Johan Vikstrom [Fri, 30 Aug 2019 09:33:27 +0000 (09:33 +0000)]
[clangd] Collecting main file macro expansion locations in ParsedAST.

Summary: TokenBuffer does not collect macro expansions inside macro arguments which is needed for semantic higlighting. Therefore collects macro expansions in the main file in a PPCallback when building the ParsedAST instead.

Reviewers: hokein, ilya-biryukov

Subscribers: MaskRay, jkorous, arphaman, kadircet, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D66928

llvm-svn: 370452

5 years ago[Tooling] Migrated APIs that take ownership of objects to unique_ptr
Dmitri Gribenko [Fri, 30 Aug 2019 09:29:34 +0000 (09:29 +0000)]
[Tooling] Migrated APIs that take ownership of objects to unique_ptr

Subscribers: jkorous, arphaman, kadircet, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D66960

llvm-svn: 370451

5 years agodotest: improvements to the pexpect tests
Pavel Labath [Fri, 30 Aug 2019 09:07:42 +0000 (09:07 +0000)]
dotest: improvements to the pexpect tests

Summary:
While working on r370054, i've found it frustrating that the test output
was compeletely unhelpful in case of failures. Therefore I've decided to
improve that. In this I reuse the PExpectTest class, which was one of
our mechanisms for running pexpect tests, but which has gotten orhpaned
in the mean time.

I've replaced the existing send methods with a "expect" method, which
I've tried to design so that it has a similar interface to the expect
method in regular non-pexpect dotest tests (as it essentially does
something very similar). I've kept the ability to dump the transcript of
the pexpect communication to stdout in the "trace" mode, as that is a
very handy way to figure out what the test is doing. I've also removed
the "expect_string" method used in the existing tests -- I've found this
to be unhelpful because it hides the message that would be normally
displayed by the EOF exception. Although vebose, this message includes
some important information, like what strings we were searching for,
what were the last bits of lldb output, etc. I've also beefed up the
class to automatically disable the debug info test duplication, and
auto-skip tests when the host platform does not support pexpect.

This patch ports TestMultilineCompletion and TestIOHandlerCompletion to
the new class. It also deletes TestFormats as it is not testing anything
(definitely not formats) -- it was committed with the test code
commented out (r228207), and then the testing code was deleted in
r356000.

Reviewers: teemperor, JDevlieghere, davide

Subscribers: aprantl, lldb-commits

Differential Revision: https://reviews.llvm.org/D66954

llvm-svn: 370449

5 years ago[LiveDebugValues] Insert entry values after bundles
David Stenberg [Fri, 30 Aug 2019 09:06:50 +0000 (09:06 +0000)]
[LiveDebugValues] Insert entry values after bundles

Summary:
Change LiveDebugValues so that it inserts entry values after the bundle
which contains the clobbering instruction. Previously it would insert
the debug value after the bundle head using insertAfter(), breaking the
bundle.

Reviewers: djtodoro, NikolaPrica, aprantl, vsk

Reviewed By: vsk

Subscribers: hiraditya, llvm-commits

Tags: #debug-info, #llvm

Differential Revision: https://reviews.llvm.org/D66888

llvm-svn: 370448

5 years ago[clangd] Add .vscode-test to .gitignore.
Haojian Wu [Fri, 30 Aug 2019 09:06:18 +0000 (09:06 +0000)]
[clangd] Add .vscode-test to .gitignore.

Reviewers: jvikstrom

Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, kadircet, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D66949

llvm-svn: 370446

5 years ago[CodeGen]: fix error message for "=r" asm constraint
Alexander Potapenko [Fri, 30 Aug 2019 08:58:46 +0000 (08:58 +0000)]
[CodeGen]: fix error message for "=r" asm constraint

Summary:
Nico Weber reported that the following code:
  char buf[9];
  asm("" : "=r" (buf));

yields the "impossible constraint in asm: can't store struct into a register"
error message, although |buf| is not a struct (see
http://crbug.com/999160).

Make the error message more generic and add a test for it.
Also make sure other tests in x86_64-PR42672.c check for the full error
message.

Reviewers: eli.friedman, thakis

Subscribers: cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D66948

llvm-svn: 370444

5 years agovim: add `immarg` keyword
Sven van Haastregt [Fri, 30 Aug 2019 08:52:55 +0000 (08:52 +0000)]
vim: add `immarg` keyword

The `immarg` attribute was added in r355981.

llvm-svn: 370443

5 years agogn build: Merge r370441
Nico Weber [Fri, 30 Aug 2019 08:26:37 +0000 (08:26 +0000)]
gn build: Merge r370441

llvm-svn: 370442

5 years ago[ADT] Removed VariadicFunction
Dmitri Gribenko [Fri, 30 Aug 2019 08:21:55 +0000 (08:21 +0000)]
[ADT] Removed VariadicFunction

Summary:
It is not used. It uses macro-based unrolling instead of variadic
templates, so it is not idiomatic anymore, and therefore it is a
questionable API to keep "just in case".

Subscribers: mgorny, dmgreen, dexonsmith, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D66961

llvm-svn: 370441

5 years ago[lldb][NFC] Move Clang-specific flags to ClangUserExpression
Raphael Isemann [Fri, 30 Aug 2019 07:44:29 +0000 (07:44 +0000)]
[lldb][NFC] Move Clang-specific flags to ClangUserExpression

LLVMUserExpression doesn't use these variables and they are all specific to Clang.

Also removes m_const_object as this was actually never used by anyone (and Clang
didn't report it as we assigned it in the constructor which seems to count as use).

llvm-svn: 370440

5 years ago[ELF] Set `referenced` bit of Undefined created by BitcodeFile
Fangrui Song [Fri, 30 Aug 2019 07:10:30 +0000 (07:10 +0000)]
[ELF] Set `referenced` bit of Undefined created by BitcodeFile

D64136 and D65584, while fixing STB_WEAK issues and improving our
compatibility with ld.bfd, can cause another STB_WEAK problem related to
LTO:

If %tundef.o has an undefined reference on f,
and %tweakundef.o has a weak undefined reference on f,
%tdef.o has a definition of f

```
ld.lld %tundef.o %tweakundef.o --start-lib %tdef.o --end-lib
```

1) `%tundef.o` doesn't set the `referenced` bit.
2) `%weakundef.o` changes the binding from STB_GLOBAL to STB_WEAK
3) `%tdef.o` is not fetched because the binding is weak.

Step (1) is incorrect. This patch sets the `referenced` bit of Undefined
created by bitcode files.

Reviewed By: ruiu

Differential Revision: https://reviews.llvm.org/D66992

llvm-svn: 370437

5 years ago[LLD] [COFF] Support merging resource object files
Martin Storsjo [Fri, 30 Aug 2019 06:56:33 +0000 (06:56 +0000)]
[LLD] [COFF] Support merging resource object files

Extend WindowsResourceParser to support using a ResourceSectionRef for
loading resources from an object file.

Only allow merging resource object files in mingw mode; keep the
existing error on multiple resource objects in link mode.

If there only is one resource object file and no .res resources,
don't parse and recreate the .rsrc section, but just link it in without
inspecting it. This allows users to produce any .rsrc section (outside
of what the parser supports), just like before. (I don't have a specific
need for this, but it reduces the risk of this new feature.)

Separate out the .rsrc section chunks in InputFiles.cpp, and only include
them in the list of section chunks to link if we've determined that there
only was one single resource object. (We need to keep other chunks from
those object files, as they can legitimately contain other sections as
well, in addition to .rsrc section chunks.)

Differential Revision: https://reviews.llvm.org/D66824

llvm-svn: 370436

5 years ago[WindowsResource] Remove use of global variables in WindowsResourceParser
Martin Storsjo [Fri, 30 Aug 2019 06:56:02 +0000 (06:56 +0000)]
[WindowsResource] Remove use of global variables in WindowsResourceParser

Instead of updating a global variable counter for the next index of
strings and data blobs, pass along a reference to actual data/string
vectors and let the TreeNode insertion methods add their data/strings to
the vectors when a new entry is needed.

Additionally, if the resource tree had duplicates, that were ignored
with -force:multipleres in lld, we no longer store all versions of the
duplicated resource data, now we only keep the one that actually ends
up referenced.

Differential Revision: https://reviews.llvm.org/D66823

llvm-svn: 370435

5 years ago[WindowsResource] Avoid duplicating the input filenames for each resource. NFC.
Martin Storsjo [Fri, 30 Aug 2019 06:55:54 +0000 (06:55 +0000)]
[WindowsResource] Avoid duplicating the input filenames for each resource. NFC.

Differential Revision: https://reviews.llvm.org/D66821

llvm-svn: 370434

5 years ago[COFF] Add a ResourceSectionRef method for getting resource contents
Martin Storsjo [Fri, 30 Aug 2019 06:55:49 +0000 (06:55 +0000)]
[COFF] Add a ResourceSectionRef method for getting resource contents

This allows llvm-readobj to print the contents of each resource
when printing resources from an object file or executable, like it
already does for plain .res files.

This requires providing the whole COFFObjectFile to ResourceSectionRef.

This supports both object files and executables. For executables,
the DataRVA field is used as is to look up the right section.

For object files, ideally we would need to complete linking of them
and fix up all relocations to know what the DataRVA field would end up
being. In practice, the only thing that makes sense for an RVA field
is an ADDR32NB relocation. Thus, find a relocation pointing at this
field, verify that it has the expected type, locate the symbol it
points at, look up the section the symbol points at, and read from the
right offset in that section.

This works both for GNU windres object files (which use one single
.rsrc section, with all relocations against the base of the .rsrc
section, with the original value of the DataRVA field being the
offset of the data from the beginning of the .rsrc section) and
cvtres object files (with two separate .rsrc$01 and .rsrc$02 sections,
and one symbol per data entry, with the original pre-relocated DataRVA
field being set to zero).

Differential Revision: https://reviews.llvm.org/D66820

llvm-svn: 370433

5 years ago[MIPS GlobalISel] Lower uitofp
Petar Avramovic [Fri, 30 Aug 2019 05:51:12 +0000 (05:51 +0000)]
[MIPS GlobalISel] Lower uitofp

Add custom lowering for G_UITOFP for MIPS32.

Differential Revision: https://reviews.llvm.org/D66930

llvm-svn: 370432

5 years ago[MIPS GlobalISel] Lower fptoui
Petar Avramovic [Fri, 30 Aug 2019 05:44:02 +0000 (05:44 +0000)]
[MIPS GlobalISel] Lower fptoui

Add lower for G_FPTOUI. Algorithm is similar to the SDAG version
in TargetLowering::expandFP_TO_UINT.
Lower G_FPTOUI for MIPS32.

Differential Revision: https://reviews.llvm.org/D66929

llvm-svn: 370431

5 years ago[CodeGen] Fix lowering for returning the result of an extractvalue
Dan Gohman [Fri, 30 Aug 2019 04:33:22 +0000 (04:33 +0000)]
[CodeGen] Fix lowering for returning the result of an extractvalue

When the number of return values exceeds the number of registers available,
SelectionDAGBuilder::visitRet transforms a function's return to use a
pointer to a buffer to hold return values. When the returned value is an
operator such as extractvalue, the value may have a non-zero result number.
Add that number to the indexing when obtaining the values to store.

This fixes https://bugs.llvm.org/show_bug.cgi?id=43132.

Differential Revision: https://reviews.llvm.org/D66978

llvm-svn: 370430

5 years ago[clangd] Add distinct highlightings for static fields and methods
Nathan Ridge [Fri, 30 Aug 2019 03:37:24 +0000 (03:37 +0000)]
[clangd] Add distinct highlightings for static fields and methods

Reviewers: hokein, ilya-biryukov, jvikstrom

Reviewed By: hokein

Subscribers: MaskRay, jkorous, arphaman, kadircet, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D66828

llvm-svn: 370429

5 years ago[PowerPC][NFC] Use inline Subtarget->isPPC64()
Jinsong Ji [Fri, 30 Aug 2019 03:16:41 +0000 (03:16 +0000)]
[PowerPC][NFC] Use inline Subtarget->isPPC64()

To be consistent with all the other instances.

llvm-svn: 370428

5 years ago[PowerPC][NFC] Use -mtriple in RUN line, remove target triple in tls.ll
Jinsong Ji [Fri, 30 Aug 2019 02:57:33 +0000 (02:57 +0000)]
[PowerPC][NFC] Use -mtriple in RUN line, remove target triple in tls.ll

To avoid confusion, especially when -mtriple are also added for PPC32.

llvm-svn: 370427

5 years ago[PPC32] Emit R_PPC_GOT_TPREL16 instead R_PPC_GOT_TPREL16_LO
Fangrui Song [Fri, 30 Aug 2019 02:20:49 +0000 (02:20 +0000)]
[PPC32] Emit R_PPC_GOT_TPREL16 instead R_PPC_GOT_TPREL16_LO

Unlike ppc64, which has ADDISgotTprelHA+LDgotTprelL pairs,
ppc32 just uses LDgotTprelL32, so it does not make lots of sense to use
_LO without a paired _HA.

Emit R_PPC_GOT_TPREL16 instead R_PPC_GOT_TPREL16_LO to match GCC, and
get better linker relocation check. Note, R_PPC_GOT_TPREL16_{HA,LO}
don't have good linker support:

(a) lld does not support R_PPC_GOT_TPREL16_{HA,LO}.
(b) Top of tree ld.bfd does not support R_PPC_GOT_REL16_HA Initial-Exec -> Local-Exec relaxation:

  // a.o
  addis 3, 3, tsd_tls@got@tprel@ha
  lwz 3, tsd_tls@got@tprel@l(3)
  add 3, 3, tsd_tls@tls
  // b.o
  .section .tdata,"awT"; .globl tsd_tls; tsd_tls:

  // ld/ld-new a.o b.o
  internal error, aborting at ../../bfd/elf32-ppc.c:7952 in ppc_elf_relocate_section

Reviewed By: adalava

Differential Revision: https://reviews.llvm.org/D66925

llvm-svn: 370426

5 years ago[clang-scan-deps] NFC, refactor the DependencyScanningWorker to use a consumer
Alex Lorenz [Fri, 30 Aug 2019 01:25:57 +0000 (01:25 +0000)]
[clang-scan-deps] NFC, refactor the DependencyScanningWorker to use a consumer
to report the dependencies to the client

This will allow the scanner to report modular dependencies to the consumer.
This will also allow the scanner to accept regular cc1 clang invocations, e.g.
in an implementation of a libclang C API for clang-scan-deps, that I will add
follow-up patches for in the future.

llvm-svn: 370425

5 years ago[X86] Explicitly list all the always trivially rematerializable instructions.
Craig Topper [Fri, 30 Aug 2019 00:54:36 +0000 (00:54 +0000)]
[X86] Explicitly list all the always trivially rematerializable instructions.

Add a default with an llvm_unreachable for anything we don't expect.

This seems safer that just blindly returning true for anything
missing from the switch.

llvm-svn: 370424

5 years agoDebugInfo: add CodeView register mapping for ARM NT
Saleem Abdulrasool [Fri, 30 Aug 2019 00:16:02 +0000 (00:16 +0000)]
DebugInfo: add CodeView register mapping for ARM NT

Add the core registers and NEON registers mapping to the CodeView
register ID.  This is sufficient to compile a basic C program with debug
info using CodeView debug info.

llvm-svn: 370423

5 years ago[Modules] Make ReadModuleMapFileBlock errors reliable
Bruno Cardoso Lopes [Thu, 29 Aug 2019 23:14:08 +0000 (23:14 +0000)]
[Modules] Make ReadModuleMapFileBlock errors reliable

This prevents a crash when an error should be emitted instead.

During implicit module builds, there are cases where ReadASTCore is called with
ImportedBy set to nullptr, which breaks expectations in ReadModuleMapFileBlock,
leading to crashes.

Fix this by improving ReadModuleMapFileBlock to handle ImportedBy correctly.
This only happens non deterministically in the wild, when the underlying file
system changes while concurrent compiler invocations use implicit modules,
forcing rebuilds which see an inconsistent filesystem state. That said, there's
no much to do w.r.t. writing tests here.

rdar://problem/48828801

llvm-svn: 370422

5 years ago[CMake][Fuchsia] Enable experimental pass manager by default
Petr Hosek [Thu, 29 Aug 2019 23:12:06 +0000 (23:12 +0000)]
[CMake][Fuchsia] Enable experimental pass manager by default

We plan on using experimental new pass manager for Fuchsia toolchain.

Differential Revision: https://reviews.llvm.org/D58214

llvm-svn: 370421

5 years ago[clang-scan-deps] reuse the file manager across invocations of
Alex Lorenz [Thu, 29 Aug 2019 22:56:38 +0000 (22:56 +0000)]
[clang-scan-deps] reuse the file manager across invocations of
the dependency scanner on a single worker thread

This behavior can be controlled using the new `-reuse-filemanager` clang-scan-deps
option. By default the file manager is reused.

The added test/ClangScanDeps/symlink.cpp is able to pass with
the reused filemanager after the related FileEntryRef changes
landed earlier. The test test/ClangScanDeps/subframework_header_dir_symlink.m
still fails when the file manager is reused (I run the FileCheck with not to
make it PASS). I will address this in a follow-up patch that improves
the DirectoryEntry name modelling in the FileManager.

llvm-svn: 370420

5 years agoFix silent wrong-code bugs and crashes with designated initialization.
Richard Smith [Thu, 29 Aug 2019 22:49:34 +0000 (22:49 +0000)]
Fix silent wrong-code bugs and crashes with designated initialization.

We failed to correctly handle the 'holes' left behind by designated
initializers in VerifyOnly mode. This would result in us thinking that a
designated initialization would be valid, only to find that it is not
actually valid when we come to build it. In a +Asserts build, that would
assert, and in a -Asserts build, that would silently lose some part of
the initialization or crash.

With this change, when an InitListExpr contains any designators, we now
always build a structured list so that we can track the locations of the
'holes' that we need to go back and fill in.

We could in principle do better: we only need the structured form if
there is a designator that jumps backwards (and can otherwise check for
the holes as we progress through the initializer list), but dealing with
that turns out to be rather complicated, so it's not done as part of
this patch.

llvm-svn: 370419

5 years agoRefactor InitListChecker to check only a single (explicit) initializer
Richard Smith [Thu, 29 Aug 2019 22:49:33 +0000 (22:49 +0000)]
Refactor InitListChecker to check only a single (explicit) initializer
list, rather than recursively checking multiple lists in C.

This simplification is in preparation for making InitListChecker
maintain more state that's specific to the explicit initializer list,
particularly when handling designated initialization.

llvm-svn: 370418

5 years agoRefactor InitListChecker to make it a bit clearer that hasError is only
Richard Smith [Thu, 29 Aug 2019 22:49:32 +0000 (22:49 +0000)]
Refactor InitListChecker to make it a bit clearer that hasError is only
set to true in VerifyOnly mode in cases where it's also set to true when
actually building the initializer list.

Add FIXMEs for the two cases where that's not true. No functionality
change intended.

llvm-svn: 370417

5 years ago[WebAssembly] Implement NO_STRIP
Dan Gohman [Thu, 29 Aug 2019 22:41:05 +0000 (22:41 +0000)]
[WebAssembly] Implement NO_STRIP

This patch implements support for the NO_STRIP flag, which will allow
__attribute__((used)) to be implemented.

This accompanies https://reviews.llvm.org/D62542, which moves to setting the
NO_STRIP flag, and will continue to set EXPORTED for Emscripten targets for
compatibility.

Differential Revision: https://reviews.llvm.org/D66968

llvm-svn: 370416

5 years ago[WebAssembly] Make __attribute__((used)) not imply export.
Dan Gohman [Thu, 29 Aug 2019 22:40:00 +0000 (22:40 +0000)]
[WebAssembly] Make __attribute__((used)) not imply export.

Add an WASM_SYMBOL_NO_STRIP flag, so that __attribute__((used)) doesn't
need to imply exporting. When targeting Emscripten, have
WASM_SYMBOL_NO_STRIP imply exporting.

Differential Revision: https://reviews.llvm.org/D62542

llvm-svn: 370415

5 years ago[Tests] Precommit a few cases where we're missing oppurtunities for block local simpl...
Philip Reames [Thu, 29 Aug 2019 22:08:17 +0000 (22:08 +0000)]
[Tests] Precommit a few cases where we're missing oppurtunities for block local simplications off assumes.

llvm-svn: 370414

5 years ago[lit] Print exit code in for unresolved (lldb)tests.
Jonas Devlieghere [Thu, 29 Aug 2019 22:02:28 +0000 (22:02 +0000)]
[lit] Print exit code in for unresolved (lldb)tests.

A test is marked unresolved when we're unable to find PASSED or FAILED
in the dotest output. Usually this is because we crashed and when that
happens the exit code can give a clue as to why. This patch adds the
exit code to the lit output to make it easier to investigate those
issues.

Differential revision: https://reviews.llvm.org/D66975

llvm-svn: 370413

5 years ago[NFC] Test commit - sorted headers.
Nandor Licker [Thu, 29 Aug 2019 21:57:47 +0000 (21:57 +0000)]
[NFC] Test commit - sorted headers.

llvm-svn: 370412

5 years ago[PowerPC] Support extended mnemonics mffprwz etc.
Jinsong Ji [Thu, 29 Aug 2019 21:53:59 +0000 (21:53 +0000)]
[PowerPC] Support extended mnemonics mffprwz etc.

Summary:
Reported in https://github.com/opencv/opencv/issues/15413.

We have serveral extended mnemonics for Move To/From Vector-Scalar Register Instructions
eg: mffprd,mtfprd etc.

We only support one of them, this patch add the others.

Reviewers: nemanjai, steven.zhang, hfinkel, #powerpc

Reviewed By: hfinkel

Subscribers: wuzish, qcolombet, hiraditya, kbarton, MaskRay, shchenz, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D66963

llvm-svn: 370411

5 years ago[AArch64][GlobalISel] Select arithmetic extended register patterns
Jessica Paquette [Thu, 29 Aug 2019 21:53:58 +0000 (21:53 +0000)]
[AArch64][GlobalISel] Select arithmetic extended register patterns

This teaches GISel to select patterns which fold an extend plus optional shift
into the addressing mode. In particular, adds and subs.

Factor out the arith extended register ComplexPatterns in AArch64InstrFormats.td
and create GISel equivalents.

Add some equivalent functions to the ones in AArch64ISelDAGToDAG:

- `selectArithExtendedRegister`
- `narrowExtendRegIfNeeded`
- `getExtendTypeForInst`

`getExtendTypeForInst` includes the checks for loads and stores. This will be
used for WRO addressing modes in loads + stores.

Teach selectCopy to properly handle subregister copies on the same bank in
order to support `narrowExtendRegIfNeeded`. The extended register must be a
GPR32, so we need to support same-bank subregister copies.

Fix a bug in getSubRegForClass which would cause registers on things like
GPR32common to end up getting ssub. Just change the check to look for FPR32
rather than GPR32.

For tests:

- Add select-arith-extended-reg.mir
- Update addsub_ext.ll to include GlobalISel checks

Differential Revision: https://reviews.llvm.org/D66835

llvm-svn: 370410

5 years ago[X86] Don't emit unreachable stack adjustments
Reid Kleckner [Thu, 29 Aug 2019 21:24:41 +0000 (21:24 +0000)]
[X86] Don't emit unreachable stack adjustments

Summary:
This is a minor improvement on our past attempts to do this. Fixes
PR43155.

Reviewers: hans

Subscribers: hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D66905

llvm-svn: 370409

5 years agoAllow '@' to appear in x86 mingw symbols
Reid Kleckner [Thu, 29 Aug 2019 21:15:02 +0000 (21:15 +0000)]
Allow '@' to appear in x86 mingw symbols

Summary:
There is no reason to differ in assembler behavior here between -msvc
and -gnu targets. Without this setting, the text after the '@' is
interpreted as a symbol variable, like foo@IMGREL.

Reviewers: mstorsjo

Subscribers: hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D66974

llvm-svn: 370408

5 years ago[InstCombine] add possible bswap as widening shuffle test; NFC
Sanjay Patel [Thu, 29 Aug 2019 20:57:50 +0000 (20:57 +0000)]
[InstCombine] add possible bswap as widening shuffle test; NFC

Goes with the proposal in D66965.

llvm-svn: 370407

5 years ago[CFG] Fix CFG for statement-expressions in return values.
Artem Dergachev [Thu, 29 Aug 2019 20:37:28 +0000 (20:37 +0000)]
[CFG] Fix CFG for statement-expressions in return values.

We're building the CFG from bottom to top, so when the return-value expression
has a non-trivial CFG on its own, we need to continue building from the entry
to the return-value expression CFG rather than from the block to which
we've just appended the return statement.

Fixes a false positive warning "control may reach end of non-void function".

llvm-svn: 370406